Frequency filter bank for enhancing carbon nanotube images

Abstract

Improving digital images of carbon nanotubes is an important task for characterizing nanotube structures in Nanoscience and Nanotechnology. A two-step algorithm is proposed for enhancing the information of carbon nanotube images, which are obtained by a scanning electron microscopy. In the first step it is carried out the characterization of the intensity profile of the nanotube by using the first and second derivatives along with the local variance. Then, for analyzing the intensity profile of the nanotubes, an adaptive spatial filter is designed. The first step allows to represent the intensity profile of the nanotube through a Gaussian model. In the second step, a Gaussian-matched filter Bank is designed in the frequency domain for enhancing the nanotube information, considering different values of thickness and orientation for the filter bank.